Certified Legendary Thread The Squiggle is back in 2023 (and other analytics)

[Final Siren]

If and it’s a big IF 17th or 16th are bad enough I can north winning it’s way out of a spoon.
There just needs to be a small gap in the last 8-9 rounds IMO.
Once Tarrant, Corr, McDonald, Anderson, Polec, Dumont, Hall etc come back after the bye we’ll see a sharp increase in competitiveness IMO

Yep, and the thing with the spoon is you almost always need just a brief patch of good form to avoid it. As recently as last year, Adelaide looked like the surest of sure things to finish bottom, then almost escaped over the last four games:

After Round 14
17th North 3 wins 11 losses @ 76%
18th Adelaide 0 wins 13 losses @ 55%

End of Season
17th North 3 wins 14 losses @ 71%
18th Adelaide 3 wins 14 losses @ 64%

The other thing is that teams can get sharply worse once they decide the season is a write-off and it's time to look to the future, booking in players for surgery or picking more kids. So a team that's only kind of bad at the moment might become seriously bad later, and be overtaken by an improving North.

 

[Mantastic]

Squiggle 1 point off a perfect tip in the Cats/Suns game.

 

[BEaston]

that bulldogs movement

 

[SuperVolley11]

You can see how the game has changed from the mid 00s and now the late 10s/20. With the exception of 3 or 4 freak sides right at the top

 

[Boristown]

that bulldogs movement

Squiggle has a hard on for the Dogs.

 

[Final Siren]

Squiggle 1 point off a perfect tip in the Cats/Suns game.

Got Freo v Sydney exactly right, too! These things are mostly luck - you would rather be reliably close than occasionally bang-on and other times way off - but I get excited about them anyway.

 

[Joao]

Got Freo v Sydney exactly right, too! These things are mostly luck - you would rather be reliably close than occasionally bang-on and other times way off - but I get excited about them anyway.

Did squiggle tip our terrible goal kicking? I saw the score was right but given we moved it must have thought we'd kick better (which is laughable because we are dire in front of the sticks).

 

[Final Siren]

Did squiggle tip our terrible goal kicking? I saw the score was right but given we moved it must have thought we'd kick better (which is laughable because we are dire in front of the sticks).

Squiggle expected a 75-73 win and in reality it was 86-84, so both teams scored a bit more than expected. This will often result in both teams moving up and left on the chart (increasing attack rating, decreasing defence rating), but Sydney's score gets discounted a bit because their 13.6 (84) reflects unusually accurate conversion, which isn't sustainable. No adjustment for Freo's score, since 12.14 isn't especially remarkable.

Therefore Squiggle banks the correct tip & margin, but internally makes these adjustments:

• Freo's attack is a bit better than previously thought, because they scored more highly than tipped
• Freo's defence is as expected, because Sydney would have produced their tipped score with more typical conversion
• Sydney's attack is as expected, for the same reason as above
• Sydney's defence is a bit worse than previously thought, since Freo scored more highly than tipped

 

[Boristown]

How will the Squiggle handle all these fixture changes?

 

[Dirty Bird]

Im actually more intrigued to see how Squiggle handles the lack of HGA with no fans in the 3 Vic games which havnt been moved.

 

[Final Siren]

How will the Squiggle handle all these fixture changes?

We got through 2020; this is nothing.

Im actually more intrigued to see how Squiggle handles the lack of HGA with no fans in the 3 Vic games which havnt been moved.

Mmmm ok this one requires a manual update! I don't have a bot that automatically fetches COVID restrictions.

 

[roacheee]

The dogs squiggle is way off

 

[Final Siren]

The dogs squiggle is way off

I'm guessing this is related to the common idea that Squiggle overrates thrashings against bad teams (e.g. the Dogs' 111-pt defeat of St Kilda) and underrates performances against good teams (e.g. Dogs' 28-pt loss to Melbourne).

Or, from Squiggle's perspective, humans focus too much on a small sample of results that are easily influenced by randomness, and are too quick to dismiss emphatic proof of high performance.

This comes up a lot and I'm sure there are examples where each view has been proved more correct than the other. It's also possible that Squiggle's approach is better suited to rating team strength in general, which is what it really cares about, while the human approach is more suited to predicting the results of key games in particular, like Grand Finals, which is what humans really care about.

 

[Crankyhawk]

The main page is the Aggregate Ladder, i.e. this: https://squiggle.com.au/ladder/

That combines the predictions of a dozen or so good models from around the internet, including mine, and that process smooths out the little quirks of each one. It should be a more reliable prediction than that of any model in particular.

Although, it just occurs to me, I haven't actually tested that. And (flex warning), last year Squiggle's in-house ladder prediction, which you see on the Live Squiggle page, was more accurate than anyone else's (via Ladder Scoreboard).

As far as I know, no-one has comprehensively measured ladder prediction accuracy before. So I don't think there's been the same attention or accountability as for game tips, nor a similar motivation to get better at it.

Bottom-line, I would rely on the Aggregate Ladder rather than any other single model's prediction (even mine), but remain aware that there's still plenty of room for improvement.


I don't have a good way of quickly checking, because it requires running tens of thousands of simulations per round. But I'm sure it's this year, since North are extremely bad, even by historical standards, and whoever finishes 2nd-last is probably going to be better than usual for that ladder position... maybe a lot better than usual.

So there is a gap there that didn't exist for Melbourne 2013, Fitzroy 1996, et al.

Hawks will be around norths mark

 

[Final Siren]

Hawks will be around norths mark

Still a 2-game + percentage gap between them on projections, although that will change very quickly if North hit a patch of good form.

Hawks do look destined for the bottom 2 (80% likelihood), because the only other team who might drop in there is Collingwood, who aren't that bad and don't even have a 1st-round draft pick to (not) play for.

 

[Crankyhawk]

Still a 2-game + percentage gap between them on projections, although that will change very quickly if North hit a patch of good form.

Hawks do look destined for the bottom 2 (80% likelihood), because the only other team who might drop in there is Collingwood, who aren't that bad and don't even have a 1st-round draft pick to (not) play for.

View attachment 1142184

we are heading towards them and I reckon we will get there (on the squiggle chart)

 

[Ocha905]

Bulldogs are looking like they're in a different class this year.

 

[Billy ray]

Oh hai everyone

Nice movement by the SUNS logo following tonight’s game

 

[Jeffrey Gibb]

Quick question, Mr Squiggle,

Does your time-travelling future computer-machine have the ability to prematurely/posthumourously assess the relative ease/difficulty of every team’s fixture before and after each season?

(Made waaaay more difficult by CoVID-base venue changes, I would assume, especially because of HGA, mRNA, LAFHA, PCP, DDT, SaaS and TB4.)

If so, would be super interesting (and helpful for trolling and excuse-making.)

 

[Final Siren]

Quick question, Mr Squiggle,

Does your time-travelling future computer-machine have the ability to prematurely/posthumourously assess the relative ease/difficulty of every team’s fixture before and after each season?

(Made waaaay more difficult by CoVID-base venue changes, I would assume, especially because of HGA, mRNA, LAFHA, PCP, DDT, SaaS and TB4.)

If so, would be super interesting (and helpful for trolling and excuse-making.)

It sure does!

The problem with that analysis though is it comes in two forms:

(1) The one you do before the season, which everyone is very interested in, but because it relies on assumptions about who will have strong/weak double-up games, tends to be pretty wrong

(2) The one you do after the season, which is accurate but nobody cares because GOD the season is over, get over it

Recent years have been particularly bad for #1 because we can't even get the venues right in advance.

 

[Vindicater]

Everytime Icome back to this thread I remember how much of a legend you are Final Siren

 

[Jeffrey Gibb]

It sure does!

The problem with that analysis though is it comes in two forms:

(1) The one you do before the season, which everyone is very interested in, but because it relies on assumptions about who will have strong/weak double-up games, tends to be pretty wrong

(2) The one you do after the season, which is accurate but nobody cares because GOD the season is over, get over it

Recent years have been particularly bad for #1 because we can't even get the venues right in advance.

Thanks for the reply.

It would be a moveable feast, no doubt.

It may show how a fixture remained consistently difficult through a season. Or how pre-season prognostications were wildly off and a difficult fixture turned into a doddle.

But it could explain how a team might yo-yo from one year to the next. I guess i’m most interested to see whether it reflects the AFL’s inconsistently cack-handed attempts at equalisation. For example, can an easy draw ‘drag’ a team up the ladder, or do they bomb out in finals anyway. Or, does a premiers difficult draw bring them back to the pack, or makes no difference to finals results as they’re better prepared against better opposition.

(Or maybe we all just need another source of confirmation bias. I know I do. Our draw seems tough even though we suuuck.)

Shrugs. Thanks anyhow. Appreciate your hard work. Love the Squiggle.

 

[Final Siren]

Everytime Icome back to this thread I remember how much of a legend you are Final Siren

Now this is the kind of content I come here for

 

[Final Siren]

Thanks for the reply.

It would be a moveable feast, no doubt.

It may show how a fixture remained consistently difficult through a season. Or how pre-season prognostications were wildly off and a difficult fixture turned into a doddle.

But it could explain how a team might yo-yo from one year to the next. I guess i’m most interested to see whether it reflects the AFL’s inconsistently cack-handed attempts at equalisation. For example, can an easy draw ‘drag’ a team up the ladder, or do they bomb out in finals anyway. Or, does a premiers difficult draw bring them back to the pack, or makes no difference to finals results as they’re better prepared against better opposition.

(Or maybe we all just need another source of confirmation bias. I know I do. Our draw seems tough even though we suuuck.)

Shrugs. Thanks anyhow. Appreciate your hard work. Love the Squiggle.

We can look at a particular year you're interested in, if you like.

Broadly speaking, though...

There can be a pretty big difference between a team's actual ladder position and the one they "deserved," in a totally fair universe. One of the best ways to predict who will rise or fall the next year is just to look at who got lucky/unlucky this year and expect them to finish around where they actually deserved, because that comes true quite often.

But mostly this is due to the results of close games, which are not won by better teams so much as luckier ones.

(Horizontal line means no correlation. )

It's actually pretty rare to find a close game that you can say was won by a team off the back of an undeserved home advantage or undeservedly difficult double-up game. So while we can calculate that a particularly good fixture should be worth just under 1 win per season on average, a lot of the time, a team won't actually get that 1 win. And a team with a bad fixture might fall over the line in a couple of games and finish higher than they really deserved.

Even in 2020, which served up a historically unbalanced fixture due to venue changes, it didn't have a huge impact on results, and the Grand Final was played between two disadvantaged teams.

So while it's definitely true that some teams have a leg up on others because of fixturing, it's not as big of an issue as is often made out. Instead of having a great fixture, it's better to be good and lucky.

 

[Crankyhawk]

We can look at a particular year you're interested in, if you like.

Broadly speaking, though...

There can be a pretty big difference between a team's actual ladder position and the one they "deserved," in a totally fair universe. One of the best ways to predict who will rise or fall the next year is just to look at who got lucky/unlucky this year and expect them to finish around where they actually deserved, because that comes true quite often.

But mostly this is due to the results of close games, which are not won by better teams so much as luckier ones.

(Horizontal line means no correlation. )

It's actually pretty rare to find a close game that you can say was won by a team off the back of an undeserved home advantage or undeservedly difficult double-up game. So while we can calculate that a particularly good fixture should be worth just under 1 win per season on average, a lot of the time, a team won't actually get that 1 win. And a team with a bad fixture might fall over the line in a couple of games and finish higher than they really deserved.

Even in 2020, which served up a historically unbalanced fixture due to venue changes, it didn't have a huge impact on results, and the Grand Final was played between two disadvantaged teams.

So while it's definitely true that some teams have a leg up on others because of fixturing, it's not as big of an issue as is often made out. Instead of having a great fixture, it's better to be good and lucky.

In that attachment any reason why hawks 2016 is noted?